PaperPulse AI Cover Art

PaperPulse AI

Daily AI research digest. Top arXiv papers explained in plain English by Alex and Sam.

Episodes
Episode 2 • Latest

#2: Objective vs. Search: Decomposing Tokenisers + Zeroth-Order Alignment

Standards: EBU R128 (-16 LUFS) • 128 kbps CBR • 7.07 MB

In this episode, Alex and Sam unpack why search procedure dominates over compression objective in tokenizer quality, and explore ComPO: zeroth-order LLM preference optimization without backpropagating through preference loss.

Episode 1

#1: Pilot: Untangling Tokenizers & Zeroth-Order LLM Alignment

Duration: 4m 05s • Standards: EBU R128 (-16 LUFS) • 128 kbps CBR • 3.74 MB

The pilot episode introducing Alex and Sam breaking down the newest pre-prints from arXiv categories cs.AI, cs.LG, and cs.CL for commuters.